A suitable vegetation index for quantifying temporal variation of leaf area index (LAI) in semiarid mixed grassland

Determining a suitable vegetation index (VI) is important for accurately quantifying temporal variation in leaf area index (LAI) in semiarid mixed grassland where performances of VI are highly influenced by complex canopy composition, large amounts of dead vegetation, and bare soil. This study investigates the effects of canopy composition on LAI, evaluates the performances of 16 hyperspectral VIs on LAI estimation at each growing stage, and documents the sensitivity of the VIs to green vegetation and the resistance to dead vegetation and bare soil. Finally, the most suitable VI for quantifying LAI temporal variation is determined. The results indicate that dead vegetation accounts for more variation in LAI than does green vegetation. The study also shows that performances of ratio-based, chlorophyll-independent, soil-line-litter-corrected (L-ATSVI), and soil-line-related VIs are more reliable and stable than chlorophyll-corrected VIs and atmospheric resistant vegetation index (SARVI). Normalized difference vegetation index (NDVI), standardized LAI determining index (SLAIDI), and transformed soil-adjusted vegetation index (TSAVI) are the optimum VIs for LAI estimations in the early, maximum, and late growing seasons, respectively. NDVI is the most suitable VI for quantifying temporal variation in LAI during the entire growing season.